Re: [idn] draft-ietf-idn-requirements-04.txt (revised copy)

[Frank Ernens <fgernens@enternet.com.au>]

OLD:

> [25] Case folding MUST be locale independent. For example, Latin
> capital letter I (U+0049) case folded to lower case in the Turkish
> context will become Latin small letter dotless i (U+0131). But in the
> English context, it will become Latin small letter i (U+0069).

NEW:

> [25] Case folding MUST NOT depend on the language or top-level domain.
> For example, the case of the Latin capital letter I (U+0049) folds
> identically for all languages; it does *not* have variant lower
> case forms of the Latin small letter dotless i (U+0131) or the Latin small
> letter i (U+0069) depending on whether the language is assumed to be
> Turkish or English, because the language is not known.

RATIONALE:

  i. The original text is self-contradictory. I took previous discussions
  here to mean that we would not support language tags, i.e. folding
  must be locale-independent. The original text says that but goes on
  to give an example which says exactly the opposite.
  
  ii. I have removed the word "locale" because I think it refers more
  to an artefact of some programming systems (ANSI C/POSIX etc.) than
  anything which corresponds to the real world. Traditionally it was
  the combination of country, character set and language. You will
  note that my proposed wording allows case folding to depend on
  some character set tag, even though I strongly believe we should
  not use those. TLD corresponds only roughly (and in some cases
  temporarily) to country.
    This was the only instance of the word "locale" in the draft.